2nd July - AI News Daily - Tech Titans Clash: Meta, Google, OpenAI & Anthropic Lead AI Revolution While Hugging Face & LangChain Reshape the Ecosystem

Send a text

Meta made significant moves by acquiring a stake in Scale AI and appointing its founder Alexandr Wang as Meta's Chief AI Officer, signaling an intensified focus on AI leadership amid company restructuring. Meanwhile, Anthropic reported $4 billion in annual recurring revenue with some executives departing to startups like Anysphere/Cursor. 

Google partnered with LangChain to launch the multimodal Gemini 2.5 researcher while YouTube expanded AI-powered search features to more U.S. users. The industry saw significant funding for startups like Genesis AI and EmeraldAI for robotics and sustainable AI infrastructure. 

New tools emerged across the ecosystem: Hugging Face introduced an Inference Providers API and now hosts 5,000+ MLX models. Inspect, an open-source AI evaluation tool trusted by Anthropic and DeepMind, is setting industry standards. Arcade's collaboration with LangChain enables secure, serverless AI integration, while LlamaCloud introduced an open-source MCP server for connecting private data with agentic applications. 

In language models, DreamPRM from CMU topped the MathVista leaderboard for multimodal reasoning. French researchers developed a 14B-parameter model outperforming larger rivals on the national Bac exam, while the MARBLE benchmark showed multimodal LLMs still struggle with complex reasoning. HuggingChat shut down after serving over a million users. 

Feature updates included Google's release of 30+ free AI education tools, YouTube's AI-driven guided search and automated summaries, and upcoming speed improvements for Gradio. 

Educational resources expanded with LangChain's new course on building AI agents with a dedicated evaluation module. Notable demos included the K-Bot affordable humanoid robot and multi-modal agents built with Gemini 2.5 that automate report and podcast creation. 

Industry discussions centered on proprietary models outpacing open-source alternatives, Stanford research showing AI models lagging behind humans in scientific idea generation, and concerns about data access as more sites block Common Crawl. Recent surveys highlighted biosecurity risks from LLMs, while industry leaders explored generative AI's impact on healthcare. 

Support the show

🌍 INAI • The Open AI Hub

The Intelligence Atlas → the world’s most comprehensive, open hub of AI knowledge. 2 Million+ tools, models, agents, tutorials & daily news—free for all, updated every day.

https://github.com/inai-sandy/inAI-wiki